- recipe deeplift
DeepLIFT (Deep Learning Important FeaTures)
- Homepage:
- Documentation:
https://github.com/kundajelab/deeplift/blob/master/README.md
- License:
OTHER / MIT License
- Recipe:
Algorithms for computing importance scores in deep neural networks.
Implements the methods in "Learning Important Features Through Propagating Activation Differences" by Shrikumar, Greenside & Kundaje, as well as other commonly-used methods such as gradients, guided backprop and integrated gradients. See https://github.com/kundajelab/deeplift for documentation and FAQ.
- package deeplift¶
- Versions:
0.6.13.0-0,0.6.12.0-0,0.6.10.0-0,0.6.9.3-0,0.6.9.1-0,0.6.9.0-0- Depends:
on numpy
>=1.9on python
- Additional platforms:
Installation¶
You need a conda-compatible package manager (currently either pixi, conda, or micromamba) and the Bioconda channel already activated (see Usage). Below, we show how to install with either pixi or conda (for micromamba and mamba, commands are essentially the same as with conda).
Pixi¶
With pixi installed and the Bioconda channel set up (see Usage), to install globally, run:
pixi global install deeplift
to add into an existing workspace instead, run:
pixi add deeplift
In the latter case, make sure to first add bioconda and conda-forge to the channels considered by the workspace:
pixi workspace channel add conda-forge
pixi workspace channel add bioconda
Conda¶
With conda installed and the Bioconda channel set up (see Usage), to install into an existing and activated environment, run:
conda install deeplift
Alternatively, to install into a new environment, run:
conda create -n envname deeplift
with envname being the name of the desired environment.
Container¶
Alternatively, every Bioconda package is available as a container image for usage with your preferred container runtime. For e.g. docker, run:
docker pull quay.io/biocontainers/deeplift:<tag>
(see deeplift/tags for valid values for <tag>).
Integrated deployment¶
Finally, note that many scientific workflow management systems directly integrate both conda and container based software deployment. Thus, workflow steps can be often directly annotated to use the package, leading to automatic deployment by the respective workflow management system, thereby improving reproducibility and transparency. Check the documentation of your workflow management system to find out about the integration.
Download stats¶
Link to this page¶
Render an 
badge with the following MarkDown:
[](http://bioconda.github.io/recipes/deeplift/README.html)
